Home
Search
Collections
Journals
About
Contact us
My IOPscience
Recruitment and blocking properties of the CardioFit stimulation lead
This article has been downloaded from IOPscience. Please scroll down to see the full text article.
2011 J. Neural Eng. 8 034004
(http://iopscience.iop.org/1741-2552/8/3/034004)
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address: 212.143.214.185
The article was downloaded on 05/05/2011 at 13:03
Please note that terms and conditions apply.
IOP PUBLISHING
JOURNAL OF NEURAL ENGINEERING
doi:10.1088/1741-2560/8/3/034004
J. Neural Eng. 8 (2011) 034004 (6pp)
Recruitment and blocking properties of
the CardioFit stimulation lead
Tamar Ahilea Anholt1 , Shai Ayal1 and Joshua A Goldberg2,3
1
BioControl Medical Ltd, Yehud 56100, Israel
Department of Physiology, Feinberg School of Medicine, Northwestern University, Chicago,
IL 60611, USA
2
E-mail: joshua-goldberg@northwestern.edu
Received 3 February 2011
Accepted for publication 22 March 2011
Published 5 May 2011
Online at stacks.iop.org/JNE/8/034004
Abstract
The CardioFit vagal stimulation system has been developed as a proposed therapy for
congestive heart failure (CHF). CardioFit is to be implanted in several hundred CHF patients
enrolled in the INOVATE-HF clinical trial, an FDA approved study. The CardioFit stimulation
lead (CSL), which is a cuff electrode that delivers stimulation pulses to the right cervical
vagus, was designed to recruit efferent cardiac vagal fibers while minimizing unwanted
recruitment of other fibers. This paper presents the CSL and measurements of its recruiting
and blocking properties when placed on isolated porcine vagus nerves maintained at an
elevated temperature in oxygenated artificial cerebrospinal fluid. Using charge balanced
quasi-trapezoidal pulses driven through the CSL, we show in eight out of nine nerves a
63% ± 13% (mean ± SD) unidirectional attenuation of the A-fiber compound action potential
attained at a current of 3.0 ± 0.8 mA. The threshold for the activation of A- and B-fibers was
found to be 0.3 ± 0.17 mA and 2.5 ± 1.1 mA, respectively. The results presented here should
help to guide the optimal parameters used in the upcoming deployment of the CardioFit
system.
in anti-epileptic VNS, is superfluous at best and could induce
unwanted side effects (Zagon and Kemeney 2000).
The CardioFit system is an implanted VNS system
that is approved for marketing in Europe and has been
implanted in about 30 patients (De Ferrari and Schwartz
2011). During INOVATE-HF (INcrease Of VAgal TonE
in chronic Heart Failure), an FDA approved clinical trial
(ClincalTrials.gov 2011), the CardioFit system will be
implanted in several hundred patients in the course of the next
few years. The main components of the CardioFit system are
the CardioFit implantable stimulator (CIS), an implantable
pulse generator; the CardioFit stimulation lead (CSL), a
nerve stimulation electrode; and an off-the-shelf intra-cardiac
sensing electrode (ISE). The CSL safety was demonstrated
using histopathological analysis of nerve samples from
chronically implanted animals (Cohen and Georgievskaya
2011).
The CSL was designed to recruit efferent vagal B-fibers,
while minimizing the recruitment of A-fibers, whose activation
could potentially have undesired central effects. The purpose
of this study was to measure the recruitment and blocking
Introduction
Vagus nerve stimulation (VNS) is an established and widely
used therapy for epilepsy (Handforth et al 1998, Heck
et al 2002) and it has also been approved as a therapy
for treatment-resistant depression (Rush et al 2005). VNS
suitability for several other applications is currently under
investigation (Castoro et al 2011). The mechanism of action
of VNS for these applications has been investigated using
various methodologies (Van Laere et al 2000), yet it is not
fully understood. It has been proposed that VNS-induced
enhancement of parasympathetic tone could serve to treat
congestive heart failure (CHF, Olshansky et al 2008, De Ferrari
and Schwartz 2011) as well as other cardiovascular diseases
(Bilgutay et al 1968, Vanoli et al 1991, Zhang et al 2002,
Tosato et al 2006). In this case, the potential therapeutic
effect presumably relies on the propagation of action potentials
along the axons of efferent cardiac vagal neurons towards the
heart. Any propagation towards the head, particularly along
A-fibers, which have been implicated in seizure suppression
3
Author to whom any correspondence should be addressed.
1741-2560/11/034004+06$33.00
1
© 2011 IOP Publishing Ltd Printed in the UK
J. Neural Eng. 8 (2011) 034004
T Ahilea Anholt et al
to the cardiac R-wave, which is sensed through the ISE. Each
burst consists of 1–3 pulses with a programmable inter-pulse
interval of 16–350 ms (typically 55 ms). Each pulse is a
quasi-trapezoidal (QT) (Accornero et al 1977, van den Honert
and Mortimer 1979, Sweeney and Mortimer 1986, Fang and
Mortimer 1991, Jones et al 1995, Tosato et al 2007), charge
balanced, asymmetric current pulse. The pulse amplitude
is programmable with current amplitudes of 0,0.1, . . . ,6 mA
and consists of a positive 500 μs plateau followed by an
approximately exponential decay with a time constant of
500 μs. This positive phase is followed by a negative discharge
phase. The electrode is allowed to discharge with a current
limited to an amplitude that is 5% of the amplitude of the
positive phase. Discharge continues until the voltage on
the electrode vanishes, which takes up to 30 ms. Because
the inter-pulse interval within a stimulation burst is longer
than the typical nerve refractory time, the neural response for
each pulse can be taken as independent of the stimulation
history. Hence, characterizing the neural response to single
pulses is the relevant information with regard to the efficacy
of the device, which delivers only low-frequency bursts.
Figure 1. Schematic drawing and electrical wiring diagram of the
CSL positioned around the nerve. Silicone is denoted in black. The
recessed anode and cathodes are denoted by A, C1 and C2,
respectively. The two indifferent contacts are denoted by a ‘0’. One
recess, marked ‘E’, is left empty (no contact) to decrease the
pressure on the nerve. The widths (in mm) are (from left to right)
0.7, 0.7, 1.4, 1.1, 0.7, and 0.7 for the recesses denoted by 0, A, C2, E,
C1, and 0, respectively. The axis of symmetry (center of the nerve)
is marked by a dashed arrow pointing in the efferent direction.
properties of the CSL as a function of stimulation current
and direction. Experiments were conducted on isolated
porcine vagus nerves. The neural response to CSL stimulation
manifested in the compound action potential (CAP) was
recorded and analyzed for relevant stimulation parameters.
CSL dimensions and materials
The CardioFit stimulation lead (CSL) (figure 1) is a modified
bipolar cuff electrode designed primarily for VNS (but it can
be applied broadly for peripheral nerve stimulation). The CSL
cuff is manufactured using injection press molding from liquid
silicone resin (NuSil, Santa Barbara, CA). Recessed into the
silicone cuff are five ring-shaped contacts made of platinum–
iridium (90%/10%). The contacts consist of one anode, two
cathodes (electrically short-circuited to each other) and two
neutral contacts (interconnected and otherwise independent
Methods
The CardioFit system
The CIS delivers stimulation bursts to the right cervical vagus
nerve using the CSL. The stimulation burst is synchronized
1
0.8
CAP Amplitude [normalized]
0.6
0.4
0.2
0
-0.2
-0.4
0
5
10
Latency [msec]
15
20
Figure 2. A typical CAP response for a 6 mA current pulse. Distinct peaks corresponding to A- and B-fibers can be seen. The A-fibers have
a double peak at a latency of less than 5 ms corresponding to the Aβ and Aγ subgroups, and the B-fibers have a peak at a latency of
9 ms. The CAP amplitude is normalized by dividing the signal with the maximal measured A-fiber CAP over all currents.
2
J. Neural Eng. 8 (2011) 034004
T Ahilea Anholt et al
of other electrodes in the cuff). The CSL’s cuff length is
12.1 mm, and its outer diameter is 5.9 mm. The CSL comes
in four models to ensure a snug fit to the nerve. The models
differ only in the cuff’s inner diameter, which can be one of
2.50 mm, 2.75 mm, 3.00 mm or 3.25 mm.
(A)
Isolated porcine vagus experimental system
Animals were supplied by, and the vagus nerve was dissected
out at, a local abattoir (Institute of Animal Research, Kibbutz
Lahav, Israel). The animals were aged 6 months of mixed
Landrace/Large White breed, and weighed approximately
90 kg. Each animal was killed by a stabbing wound to the
thorax and was bled. The vagus nerve was removed within
8 min of the stabbing and placed in an ice-cold HEPESbuffered sucrose solution containing (in mM) MgCl2 10,
CaCl2 0.5, KCl 3, glucose 11, HEPES 5, sucrose 230, pH
= 7.3–7.4 with NaOH, and shipped to our laboratory. The
nerve was then transferred into artificial cerebrospinal fluid
(ACSF) oxygenated at room temperature in 95% O2 /5% CO2 ,
which contained (in mM) MgCl2 2, CaCl2 2, KCl 2.5, NaCl
126, glucose 10, NaH2 PO4 ·H2 O 1.25 and NaHCO3 26, and
was allowed to equilibrate at room temperature for at least
75 min. After equilibration, the nerve was placed onto an
array of silver–silver chloride wires with one end in a small
‘pool’ in which the CSL was fastened around the nerve and
submerged in ACSF. The pool, containing the oxygenated
ACSF, was kept at 34 ◦ C with a feedback controller to correctly
represent the thresholds for blocking and recruitment at a
physiological temperature. The rest of the array remained
at room temperature (24 ◦ C). In all cases, a CSL with an inner
diameter of 2.5 mm was used as it provided a ‘snug’ fit to
the porcine vagus. The recording was made through a pair
of silver–silver chloride wires from the array. The recording
wires had a spacing of 5–10 mm and were at a distance of
50–90 mm from the CSL. The ground terminal was connected
to a silver–silver chloride wire, which was located between
the CSL and the recording site, and short-circuited to the bath
with a silver–silver chloride pellet. To compare activation in
the blocking direction versus the non-blocking direction of the
CSL, the CSL was removed from the nerve and its direction
was reversed.
The waveform of the pulse, generated by custom-made
software, mimicked the CardioFit pulse waveform. The
waveform was fed through a D/A board (National Instruments,
Austin, TX) to a linear stimulus isolator (WPI, Sarasota,
FL), which was connected to the CSL. The software also
controlled an A/D board that recorded 100 ms segments of
the signal. The signal was first amplified by a Cyberamp
380 (Molecular Devices, Sunnyvale, CA) amplifier (10 000×–
100 000×) in the differential configuration using a fourthorder (80 dB/decade) low-pass Bessel filter at 6 kHz for antialiasing. It was then sampled at a rate of 50 kHz. The software
enabled us to log the CAPs generated by current pulses that
varied in amplitude or duration onto a PC. The sequences
of pulses with varying parameters were delivered in random
order and multiple repetitions were averaged. The latency
and amplitude of peaks were determined offline by finding the
extrema of the waveform components recorded.
10
20
8
15
6
10
4
Stimulation Current [mA]
2
5
Latency [msec]
00
(B)
10
20
8
15
6
10
4
Stimulation Current [mA]
2
5
Latency [msec]
00
Figure 3. (A) The CAP response for currents ranging from 0.1 to
9 mA in the non-blocking direction. The CAP amplitude is
monotonic non-decreasing as a function of current. (B) The CAP
response in the blocking direction (the same nerve as in figure 2). A
distinct region where the CAP of the A-fibers is attenuated can be
seen for currents at or above 4 mA.
Results
A typical CAP response for a 6 mA current pulse is shown
in figure 2. Distinct peaks corresponding to A- and B-fibers
may be seen. The A-fibers have a double peak (which are
frequently but not always discernable) at a latency of less than
5 ms corresponding to the Aβ and Aγ subgroups, and the
B-fibers have a peak at a latency of 9 ms. Typical CAP
responses for currents ranging from 0.1 to 9 mA in the nonblocking direction are presented in figure 3(A). The A- and
B-fiber CAP amplitudes are either rising or constant as a
function of current. In figure 3(B), we show the CAP responses
of another nerve in the blocking direction. A distinct region
where the CAP of the A-fibers (primarily the Aβ) is attenuated
can be seen in the blocking direction for currents at or above
4 mA.
The maximal CAP amplitude in the latencies relevant for
A-fibers (including both Aβ and Aγ subgroups) and B-fibers
3
J. Neural Eng. 8 (2011) 034004
T Ahilea Anholt et al
1
CAP Amplitude [normalized]
0.8
0.6
0.4
0.2
0
0
1
2
3
4
5
6
7
8
9
Stimulation Current [mA]
Figure 4. The maximal CAP amplitude in the latencies 2–6 ms relevant for A- (squares, dashed line) and 6–12 ms relevant for B-fibers
(circles, solid line) as a function of stimulation current in the same nerve as used in figures 2 and 3(B). The threshold for A- and B-fiber
activations as well as the region of the A-fiber block (marked by a black bar) can be seen. The CAP amplitude is normalized by dividing the
signal with the maximal measured A-fiber CAP over all currents. Note that this figure depicts the maximal values detected (algorithmically)
for each stimulation, resulting in positive values below threshold due to noise.
anode. Elongating the cuff would probably result in a larger
A-fiber CAP reduction, but this was impractical because it
would also increase the probability of mechanical insult to
the nerve. Previous studies of the CAP recorded from the
pig vagus in vivo (Tosato et al 2007, Vuckovic et al 2008,
Ordelman et al 2010) gave similar results with regard to
Aβ, Aγ , and B-fiber components. All of these papers give
activation thresholds for the various components of the CAP.
These thresholds coincide with our measured thresholds for
the Aβ and B CAP components. Tosato et al (2007) report
higher currents (10 mA) needed for anodal block with a QT
current pulse of the same width or wider, while maximal
block at 6.3 ± 2.3 mA was reported by Vuckovic et al
(2008) for anodal block with QT pulses of the same width
or wider, with a reduction of 60–100% of the maximal CAP
amplitude, which is comparable to what the CSL attained. In
both Tosato et al (2007) and Vuckovic et al (2008), custommade tripolar cuff electrodes were used. The quantitative
differences in percentage blocking and the blocking current
results can be attributed to the different electrode shapes and
dimensions used in these studies driving a different current
distribution through the nerve. Castoro et al (2011) measured
the excitation properties of canine vagal fibers using a helical
bipolar electrode. The activation thresholds they report for Afibers are similar to those measured here, but their measured
B-fiber threshold of 0.5 ± 0.1 mA is much lower than that
measured in the pig. Castoro et al (2011) did not report anodal
blocking in their measurements.
At 6 mA, which is the CIS’s maximal stimulation current,
a charge injection of 6 μC/phase at a charge density of
68 μC/cm2 /phase occurs at the anode (which has a smaller
as a function of stimulation current in the same recording as
shown in figure 3(B) is shown in figure 4. The threshold for
A- and B-fiber activations as well as the region of A-fiber
block (marked by a black bar) can be seen. In measurements
conducted on n = 9 nerves, we found that the threshold for
activation of A-fibers was 0.3 ± 0.17 mA (mean ± SD), while
the threshold for activation of B-fibers was 2.5 ± 1.1 mA.
In eight of the nine measured nerves, A-fiber blocking was
observed. In these fibers, maximal block of A-fibers was
attained at 3.0 ± 0.8 mA with an average reduction of 63%
± 13% from its initial peak activation, which was attained at
1.2 ± 0.4 mA. Threshold and blocking values are presented in
figure 5.
Discussion
The recruitment and blocking characteristics of the CSL were
measured on isolated vagus nerves of the pig. The results show
that the CSL, using the QT pulse produced by the CIS, is able
to excite A- and B-fibers, and that in one direction a range of
currents exists (approximately 3–5 mA) for which the A-fiber
CAP is reduced by more than 60%, while the B-fiber CAP
attains close to maximal activation. Thus, stimulation by the
CSL is B-fiber preferential and directional with regard to Afibers. Such preference toward nerve fiber type and direction
is advantageous in applications that require those features, i.e.
when organ-specific stimulation is required and the side-effect
profile needs to be minimized. A larger reduction in A-fiber
CAP was not obtained due to the virtual cathode produced
by the current flowing in the indifferent contact nearest to the
4
T Ahilea Anholt et al
60
0
40
1
50
Current [mA]
2
A−fiber max block [%]
3
70
4
J. Neural Eng. 8 (2011) 034004
A−fiber threshold
B−fiber threshold
A−fiber max block
Figure 5. Left: A- and B-fiber thresholds (n = 9) and A-fiber maximum blocking current (n = 8). Right: A-fiber CAP amplitude reduction
at maximal blocking as a percentage of the initial peak of A-fiber activation (n = 8, 100% would represent the total block).
human subjects will be similar to those measured on isolated
nerves and reported here. We should have more definitive
data about the recruitment of cardiac fibers in the human
right vagus nerve at the conclusion of the INOVATE-HF trial
(ClinicalTrials.gov 2011).
In conclusion, we show that the CSL, when driven by
the CIS, can produce a significant (>60%) reduction in the
A-fibers afferent component of the CAP. We believe that these
findings will help to guide practitioners to optimal parameter
settings to use in the stimulation of the right vagus nerve of
the prospective participants in the INOVATE-HF clinical trial
(ClinicalTrials.gov 2011).
surface area than the cathodes). While this value is above the
value used elsewhere (Shannon 1992, Popovic and Sinkjaer
2000), we note that in those studies, a stimulation frequency
of 50 Hz for a duration of 7 h was used, while CardioFit is
limited to a 9 Hz stimulation frequency at a 25% duty cycle (i.e.
stimulation is administered following at most 25% of detected
R-waves). As another safety measure, the metal contacts of
the CSL are recessed into the silicone so that the nerve does
not have mechanical contact with any rigid surface to prevent
possible tissue abrasion and damage specific to electrically
active biomaterial interfaces caused by the production of free
radicals on the electrode surface, resulting in peroxidationmediated injury to membrane lipids (Merrill et al 2005).
Indeed, chronic CardioFit stimulation for a duration of 6
months at maximal stimulation parameters was found to be
safe, not damaging the nerve fibers (Cohen and Georgievskaya
2011).
The human data for activation thresholds in the vagus
nerve for comparison to our results are scarce. In Evans
et al (2004), results from nine patients show evoked potentials
corresponding to A-, B- and C-fibers. Thresholds for
activation are not reported. In Koo et al (2001), threshold
currents for the activation and conduction velocity of vagal
fibers in humans of various ages were measured. For adults,
the threshold is below 0.5 mA, which is consistent with our
A-fiber results. The conduction velocity of adult vagal fibers
as measured in that study is 12 m s−1 , which is slower than
the 19 m s−1 measured by Evans et al (2004). Rozman
et al (2009) report influences on the heart function of cervical
left vagus stimulation in a single human subject. They show
that to elicit a significant effect, a current that is above
1.5 mA but below 2.5 mA is needed. This is consistent with our
measured threshold for activation of porcine B-fibers. While
there are some quantitative differences between our porcine
isolated nerve measurements and human measurements, the
overall picture of A- and B-fibers seems comparable. This
suggests that the recruitment characteristics of the CSL in
Acknowledgments
The authors would like to thank Rami Biran, Tsachi Czaczkes,
Itzhak Sinai, Tamir Ben-David and Ehud Cohen for excellent
technical support and valuable discussions.
References
Accornero N, Bini G, Lenzi G L and Manfredi M 1977 Selective
activation of peripheral nerve fibre groups of different diameter
by triangular shaped stimulus pulses J. Physiol 273 539–60
Bilgutay A M, Bilgutay I M, Merkel F K and Lillehei C W 1968
Vagal tuning. A new concept in the treatment of
supraventricular arrhythmias, angina pectoris and heart failure
J. Thorac. Cardiovasc. Surg. 56 71–82
Castoro M A, Yoo P B, Hincapie J G, Hamann J J, Ruble S B,
Wolf P D and Grill W M 2011 Excitation properties of the right
cervical vagus nerve in adult dogs Exp. Neurol. 227 62–8
ClinicalTrials.gov 2011
http://clinicaltrials.gov/ct2/show/NCT01303718
Cohen M L and Georgievskaya Z 2011 Histopathology of the
stimulated vagus nerve: primum non nocere Heart Fail. Rev.
16 163–9
De Ferrari G M and Schwartz P J 2011 Vagus nerve stimulation:
from pre-clinical to clinical application: challenges and future
directions Heart Fail. Rev. 16 195–203
5
J. Neural Eng. 8 (2011) 034004
T Ahilea Anholt et al
Evans M S, Verma-Ahuja S, Naritoku D K and Espinosa J A 2004
Intraoperative human vagus nerve compound action potentials
Acta Neurol. Scand. 110 232–8
Fang Z-P and Mortimer J T 1991 Selective activation of small motor
axons by quasitrapezoidal current pulses IEEE Trans. Biomed.
Eng. 38 168–74
Handforth A, DeGiorgio C M, Schachter S C, Uthman B M,
Naritoku D K, Tecoma E S, Henry T R, Collins S D,
Vaughn B V, Gilmartin R C, Labar D R, Morris G L 3rd,
Salinsky M C, Osorio I, Ristanovic R K, Labiner D M,
Jones J C, Murphy J V, Ney G C and Wheless J W 1998 Vagus
nerve stimulation therapy for partial-onset seizures: a
randomized active-control trial Neurology 51 48–55
Heck C, Helmers S L and DeGiorgio C M 2002 Vagus nerve
stimulation therapy, epilepsy, and device parameters:
scientific basis and recommendations for use Neurology
59 S31–37
Jones J F, Wang Y and Jordan D 1995 Heart rate responses to
selective stimulation of cardiac vagal C fibres in anaesthetized
cats, rats and rabbits J. Physiol. 489 203–14
Koo B, Ham S D, Sood S and Tarver B 2001 Human vagus nerve
electrophysiology: a guide to vagus nerve stimulation
parameters J. Clin. Neurophysiol. 18 429–33
Merrill D R, Bikson M and Jefferys J G 2005 Electrical stimulation
of excitable tissue: design of efficacious and safe protocols
J. Neurosci. Methods 141 171–98
Olshansky B, Sabbah H N, Hauptman P J and Colucci W S 2008
Parasympathetic nervous system and heart failure:
pathophysiology and potential implications for therapy
Circulation 118 863–71
Ordelman S C, Kornet L, Cornelussen R, Buschman H P and
Veltink P H 2010 An indirect component in the evoked
compound action potential of the vagal nerve J. Neural Eng.
7 066001
Popovic D and Sinkjaer T 2000 Control of Movement for the
Physically Disabled (London: Springer)
Rozman J, Peclin P, Knezevic I, Mirkovic T, Gersak B
and Podbregar M 2009 Heart function influenced by selective
mid-cervical left vagus nerve stimulation in a human case study
Hypertens. Res. 32 1041–3
Rush A J, Marangell L B, Sackeim H A, George M S, Brannan S K,
Davis S M, Howland R, Kling M A, Rittberg B R, Burke W J,
Rapaport M H, Zajecka J, Nierenberg A A, Husain M M,
Ginsberg D and Cooke R G 2005 Vagus nerve stimulation for
treatment-resistant depression: a randomized, controlled acute
phase trial Biol. Psychiatry 58 347–54
Shannon R V 1992 A model of safe levels for electrical stimulation
IEEE Trans. Biomed. Eng. 39 424–6
Sweeney J D and Mortimer J T 1986 An asymmetric two-electrode
cuff for generation of unidirectionally propogated action
potentials IEEE Trans. Biomed. Eng. 33 541–9
Tosato M, Yoshida K, Toft E, Nekrasas V and Struijk J J 2006
Closed-loop control of the heart rate by electrical stimulation
of the vagus nerve Med. Biol. Eng. Comput. 44 161–9
Tosato M, Yoshida K, Toft E and Struijk J J 2007 Quasi-trapezoidal
pulses to selectively block the activation of intrinsic laryngeal
muscles during vagal nerve stimulation J. Neural. Eng.
4 205–12
Van Den Honert C and Mortimer J T 1979 Generation of
unidirectionally propagated action potentials in a peripheral
nerve by brief stimuli Science 206 1311–2
Van Laere K, Vonck K, Boon P, Brans B, Vandekerckhove T
and Dierckx R 2000 Vagus nerve stimulation in refractory
epilepsy: SPECT activation study J. Nucl. Med. 41 1145–54
Vanoli E, De Ferrari G M, Stramba-Badiale H, Hull S S Jr,
Foreman R D and Schwartz P J 1991 Vagal stimulation and
prevention of sudden death in conscious dogs with a healed
myocardial infarction Circ. Res. 68 1471–81
Vuckovic A, Tosato M and Struijk J J 2008 A comparative study of
three techniques for diameter selective fiber activation in the
vagal nerve: anodal block, depolarizing prepulses and slowly
rising pulses J. Neural. Eng. 5 275–86
Zagon A and Kemeney A A 2000 Slow hyperpolarization in cortical
neurons: a possible mechanism behind vagus nerve simulation
therapy for refractory epilepsy? Epilepsia 41 1382–9
Zhang Y, Mowrey K A, Zhuang S W, Wallick D W, Popovic Z B
and Mazgalev T N 2002 Optimal ventricular rate slowing
during atrial fibrillation by feedback AV nodal-selective
vagal stimulation Am. J. Physiol. Heart Circ. Physiol.
282 H1102–10
6